European and Japanese programs to develop electric power from atomic fusion are racing ahead so much faster than American efforts that a National Research Council committee recommended yesterday that the three programs collaborate in a major international effort.

The panel, convened by this country's most prestigious scientific body, said that if it doesn't collaborate, the American scientific community could risk losing prestige and the United States could forfeit an economic advantage if and when a global market develops for fusion power plants.

The prospect of generating electricity through atomic fusion -- the welding of hydrogen atoms in the same energy-producing process that powers the sun -- has long been a dream of scientists. It is seen as an attractive alternative to the fission power used in today's atomic power plants because, unlike fission, it uses abundant fuels and produces relatively little radioactive waste.

But nobody knows whether fusion power will work.

Research began about 30 years ago in several countries with a promise that by now, fusion power would be a reality. Reaching the goal has turned out to be vastly more difficult than expected -- some say it may even be impossible -- and researchers guess that practical fusion power is perhaps 35 years away.

Scientists have yet to demonstrate that they can produce a fusion reaction in the laboratory that puts out more power than it consumes. An experimental fusion device at Princeton University was expected to be the first to do this in a test originally scheduled for 1986, but budget cuts in federal research funds have delayed the test to 1988.

Yesterday's recommendation was issued by the Committee on International Cooperation in Magnetic Fusion, a panel of scientists, engineers and officials with high-tech corporations that was convened by the National Research Council, an arm of the National Academy of Sciences.

Committee chairman Joseph G. Gavin, president of Grumman Corp., said that if the United States did not begin moving soon toward a collaborative arrangement, the other countries would have progressed so far that the United States would no longer be an attractive partner. Gavin said a "window of opportunity" for such an arrangement would open around 1990 as European and Japanese fusion programs begin to plan a new generation of experimental fusion devices.

It is only by experimenting with test devices that scientists and engineers can learn how to build more efficient reactors. There is no adequate theory of fusion to guide designers and each step is increasingly expensive. For example, it cost about $500 million to build the Princeton machine for achieving a break-even level of efficiency. A device that could sustain a fusion reaction is estimated to cost about three times as much, and a test reactor that could produce power might cost as much as $3 billion.

Although the Soviet Union is among the leaders in fusion research, Gavin said political considerations persuaded the committee that it would not be fruitful to explore Soviet-American cooperation.

Harold P. Furth, director of the Princeton fusion program but not a member of the committee, said that while collaboration was a worthy goal, the United States is perceived abroad as an unreliable partner. Several times in recent years, political motives have caused this country to withdraw from of cooperative programs

"Obviously," Furth said, "it's a good idea to pool resources with the Europeans and the Japanese, and even the Russians for that matter. But I'm not sure we have a stable enough national program to convince them that we're really committed. There's no question the Europeans and the Japanese are going ahead no matter what."